Amazon says it buys books in bulk to ‘improve products’

Amazon has revealed it purchases books in bulk through commercial channels to enhance its products and services, according to a spokesperson for the company. This move follows an investigation that traced a significant order of rare books to a facility in Las Vegas, where workers reportedly destroy and scan physical copies before scanning their pages.

The investigation, conducted by 404 Media, uncovered that approximately 1,000 books were purchased through an anonymous buyer and delivered to an Amazon facility in Las Vegas, with hidden Apple AirTags placed inside the volumes. Workers at the facility reportedly remove book spines and scan the pages, rendering the physical copies unusable.

Amazon did not specify which products the scanned books were being used to develop or confirm whether the books were being utilized to train its AI models. The company's use of printed and out-of-print books is driven by the unique content they provide, which is often not available online. Older texts also predate the surge of AI-generated material on the internet, complicating developers' efforts to source reliable human-written training data.

Amazon is developing its own family of Nova AI models and competes with tech giants such as OpenAI, Google, and Anthropic for the resources needed to create advanced AI models. The company's actions follow a similar trend among tech companies, as they scramble to acquire high-quality written material for their AI development. The demand for fresh data has become a growing challenge for AI developers, as much of the easily accessible internet has already been extensively scraped.

Publishers have capitalized on this demand, forming new revenue streams through AI licensing agreements. Amazon recently signed an AI licensing deal with The New York Times, valued between $20 million and $25 million annually.